Journal of Psycholinguistic Research

, Volume 20, Issue 2, pp 83–103

Are conjunction rule violations the result of conversational rule violations?

  • Guy Politzer
  • Ira A. Noveck

DOI: 10.1007/BF01067877

Cite this article as:
Politzer, G. & Noveck, I.A. J Psycholinguist Res (1991) 20: 83. doi:10.1007/BF01067877


This study is concerned with the conjunction error as reported by Kahneman and Tversky (1982). In the prototypical problem that manifests this error, theLinda problem, subjects are presented with a description of a 31-year-old liberal progressive named Linda and asked to rank in order a number of events concerning Linda today in terms of their likelihood. The three events of interest areLinda is a bank teller (eventA),Linda is active in the feminist movement (eventB), andLinda is a bank teller and active in the feminist movement (eventA and B). Subjects typically rankedLinda is a bank teller and active in the feminist movement (A and B) higher thanLinda is a bank teller (A). This judgment violates a fundamental law of probability, namely, thatp (A and B) ≤ p(A). Kahneman and Tversky argue that this response tendency is a result of therepresentativeness heuristic which causes subjects to ignore logical considerations of set relations and to rank the choices in terms of their resemblance to Linda's description. On the other hand, we argue that the task demands compel subjects to interpret the choiceA asA and not-B. When subjects make the unusual comparison betweenA and its subclassA and B in the context of a description of Linda that is largely irrelevant to the task's solution,Linda is a bank teller becomesLinda is a bank teller and is not active in the feminist movement. With this interpretation, subjects' typical response patterns no longer can be considered in contradiction to logical principles. One experiment is presented that uses three French languageLinda-type problems in two conditions: one in which the expected implicature is blocked and one in which the implicature is encouraged to reappear. These are both compared to a French version of theLinda problem. The structure of theLinda problem is discussed in light of conversational principles, particularly relevance (Sperber and Wilson, 1986) and the maxim of quantity (Grice, 1975).

Copyright information

© Plenum Publishing Corporation 1991

Authors and Affiliations

  • Guy Politzer
    • 1
  • Ira A. Noveck
    • 2
  1. 1.CNRS and Université de Paris VIIIParisFrance
  2. 2.Department of PsychologyNew York UniversityNew York